Every KillerBot instance is isolated, encrypted, and yours alone. We don't read your conversations. We don't train on your data. We don't share your infrastructure.
Not because we wrote a nice privacy policy. Because the architecture makes it impossible.
Every customer gets their own container. Not a row in a shared database — an actual, separate machine. Your neighbor's AI can't see your data because it doesn't exist in the same universe.
No shared databases. No shared storage. No shared compute. Your conversations, files, and agent memory live on your instance and nowhere else. We never use your data to train models.
API keys are encrypted and stored on your instance. We never have access to them. Delete your instance, everything goes with it. No backups of customer data on our side.
AES-256 encryption at rest. TLS 1.3 in transit. Firewalls locked to port 443. Dedicated Fly.io machines per customer — no shared anything. API keys encrypted on your instance. We never see them.
Learn more about infrastructureWe collect the minimum: your email for auth (via Clerk), your payment info (via Stripe). That's it. Your conversations, files, API keys, memory — all of it stays on your instance. We never use your data to train AI models. Delete your instance, it's gone.
Learn more about privacyWhen something goes wrong, we tell you. No burying it, no vague statements three weeks later. We publish what happened, how we responded, and what we changed.
View incident historyMost self-hosted AI setups are held together with duct tape and good intentions. Here's what we keep seeing:
Open instances with no authentication — anyone on the internet can walk in
API keys hardcoded in plaintext config files, committed to public repos
Shared databases with tenant IDs pretending to be "isolated"
No encryption, no monitoring, no automatic updates — just vibes
Sound familiar? Yeah, it scared us too.
Not a whitepaper. Not a checkbox exercise. Actual engineering decisions that make every KillerBot instance secure by default.
Your own container. Not a row in our database — an actual, isolated machine. Your neighbor's AI assistant literally cannot see your data because it doesn't exist in the same universe.
AES-256 at rest. TLS 1.3 in transit. Your conversations are encrypted before they hit disk and encrypted again on the way to your device. Belt and suspenders.
Dedicated compute, storage, and network per user. Not "dedicated" with air quotes — actually dedicated. Your instance runs on its own machine.
No third-party marketplace free-for-all. Remember the "ClawHavoc" campaign? 341 malicious skills found on ClawHub, siphoning API keys and conversation history. Yeah, we don't do that.
We update so you don't have to. No "please manually upgrade to fix this critical vulnerability" emails. Patches roll out automatically, zero downtime.
Your conversations never train models. Not ours, not anyone else's. We pinky-promise, and also it's in our terms of service and architecture — the data literally doesn't flow there.
No mystery boxes. Built on open-source tools and frameworks you can audit yourself. We don't hide behind proprietary walls because we don't need to.
Three users, three completely separate worlds. No shared anything.
You want specifics. We respect that. Here's the nitty-gritty.
Want to dig deeper? We're happy to talk security with your team. security@killr.bot
Common questions about how we protect your data.
7 days free. Your own isolated instance. Encrypted everything. Zero data training. We're not asking you to trust us — we're asking you to verify.
Found a vulnerability? We'd genuinely love to hear about it. security@killr.bot